Skip to main content

Overview

A dataset is a versioned table your factory relies on, such as a capacity catalog, a calendar or a code crosswalk. Its key columns identify each row, and each time new data is published it gets a new version (v1, v2, and so on). Datasets live on the Datasets tab of the Data page.

Where Datasets Come From

  • Dexter. Ask Dexter to turn an upload, or data it prepared with you, into a dataset, or to refresh one with new data.
  • Pipelines. Tables a pipeline produces are published here, with a new version whenever a run changes them.
  • New dataset. Click New dataset, choose a CSV or Excel (.xlsx) file up to 512 MB (and a Sheet if there are several), enter a Name, tick the Key columns, then click Create dataset. Rows with an empty key value must be left out with Skip those rows.

Finding and Opening a Dataset

The bar at the top of the tab shows the open dataset and branch. Click either name to switch.

The Grid

  • Search rows searches the whole dataset, not only the rows on screen.
  • Filter filters by one or more columns.
  • A column’s header menu offers Sort ascending, Sort descending, Filter, Hide column and Pin column. Drag a header to move it.
  • Columns shows and hides columns. Key columns start pinned to the left.

Editing Cells

On the Original branch, type into a cell or paste a block from a spreadsheet. A blank value clears the cell, unless the column can’t be empty. Edits don’t change the published data. They sit on top of the current version, marked with a small blue triangle in the cell’s corner, and are reapplied when a new version is published. To undo them:
  • Right-click a selection and choose Revert edit (or Revert N edits) for just those cells.
  • Click Revert edits, next to the count of edited cells, to restore every edit, or select rows first to revert only those. Confirm in the Revert edits? dialog.

Formula Columns and Number Format

Formulas adds calculated columns to the dataset. Enter a Column name and a Formula, such as rdf_cases - produced_cases, then click Add column. Formulas can use column names, numbers, + - * / %, comparisons, CASE, and abs, ceil, coalesce, floor, greatest, least, nullif and round. They recalculate when the data changes. If new data breaks one, for example by removing a column it uses, Formulas says it no longer works so you can remove it. Format sets Decimals and a Thousands separator for number columns, saved with the current view. Reset format goes back to automatic.

Branches

Every dataset has an Original branch, the copy you edit. Dexter can create other branches for what-if scenarios, such as reduced capacity. Other branches are read-only in the grid; ask Dexter to change them.

Versions and Comparing

The grid shows the latest version. Above it, Compare with vN shows how many rows changed since the previous version. Click it to compare:
  • Changed cells are blue, with the old value struck through.
  • Added rows are green, and removed rows are red.
  • New, removed and retyped columns are listed above the grid.
  • Only changes hides the rows that didn’t change.
The grid is read-only while comparing. Versions whose key columns differ can’t be compared.

Saved Views

Views are tabs above the grid; All has no saved settings. Click + to add one:
  • Save what I see keeps the current search, filters, columns and sort.
  • Ask Dexter builds a view you describe, such as “Sum the quantity by SKU and date across channels”.
The view’s … menu offers Update view and Delete. Views are saved in this factory for you and Dexter. Totals and pivot views. Dexter can build views that total rows by group (sum, average, count, minimum or maximum), optionally with a column per value of another column (a pivot). A line above the grid says what was totalled, and the Rows column counts the source rows behind each total. You can search, filter group columns, sort, hide and pin, but those changes aren’t saved. Right-click a selection and choose Color to highlight cells or matching values. Colors are saved with the view.

Notes for Dexter

Click Annotate in the sidebar (or press A), click a cell and type your note, then send it to Dexter. Dexter gets your notes in one message, each tied to its cell. Each note is also saved on its cell, so you and Dexter can read it there later.

Asking Dexter to Change a Dataset

The grid changes values only. For structural changes, ask Dexter:
  • “Refresh the Capacity dataset from the file I just uploaded”
  • “Add a Supplier column to the Lead Times dataset”
  • “Add rows for the new packing line”
  • “Create a branch of Capacity with line 3 at half capacity”
  • “Delete the reduced-capacity branch”

Limits

  • Key columns, formula columns and binary columns can’t be edited in the grid.
  • You can’t add or delete rows in the grid. Ask Dexter.
  • Branches are read-only, and edits apply only to the Original branch.
  • The grid is read-only while you compare versions or view totals.
  • A pivot shows columns for at most the first 50 values of the pivot column.
  • One change, such as a paste or reverting selected rows, can touch at most 500 rows.
  • A dataset holds at most 100 formula columns.
  • Datasets aren’t included when you share or export a factory, and rolling a factory back to an earlier version doesn’t change them.